Someone in your group chat needs a babysitter. Someone else knows a wonderful one. The introduction takes six seconds, costs nothing, and carries more weight than a page of strangers' reviews. Scott Belsky and Julio Vasconcellos saw a business in that exchange. Prefer, founded in 2015 and launched publicly in New York in 2017, set out to turn the little economy of "who do you know?" into a marketplace. Its eventual closure exposed a useful paradox: the recommendation was valuable precisely because it was personal and occasional. An app needs people to return.
- Prefer matched consumers with independent service professionals recommended by friends and phone contacts.
- Consumers paid nothing; professionals paid a 3% to 5% transaction fee.
- At launch, coverage counted more than 400 professionals in beta and a few thousand test users.
- Its most important lesson: a powerful feature may still happen too rarely to form a habit.
Four stars from strangers
The pitch was pleasingly unfashionable. The local-services market already offered a multitude of searchable profiles. Yelp and Angie's List collected reviews. Thumbtack solicited bids. Specialist apps could arrange a massage or a hair appointment. Prefer believed the first question was simpler: which professional did your friends already trust? It connected to a user's address book, surfaced people they knew on the network, and asked them to recommend providers they had actually used.
That gave Prefer two customers. The person hiring wanted confidence in someone who might enter a home, care for a child or help with an important personal task. The independent professional wanted introductions that arrived with trust attached, plus a way to manage the tedious business around the work. Prefer offered discovery and booking to consumers; a separate app gave professionals tools for schedules, payments and client relationships. It was part marketplace, part small-business operating desk.
The fee mattered. Vasconcellos said Prefer generally took 3% to 5% of transaction revenue, charging 5% when it generated new business and less for some continuing relationships. He contrasted that with open marketplaces that could charge far more. If a professional gained a repeat customer, the idea was to keep both people in the product with useful administration rather than hold them there with a tollbooth. Professionals could set their own prices and service menus, and negotiate with clients.
A fine idea with a quiet calendar
Prefer had a credible cast. Vasconcellos came from Facebook's Brazil operation and Peixe Urbano. Belsky had founded Behance and served at Benchmark. Benchmark led Prefer's Series A, with Homebrew and Expa among the backers. The team included people with experience at Twitter, Tumblr and Peixe Urbano. By May 2017, Forbes reported more than 400 professionals in beta and a few thousand test users. Those are launch figures, not evidence of later scale.
Its New York beta began in late 2016. To seed it, the founders asked people they knew to download the app, then encouraged those people to invite friends. The invitation was gentle: users were prompted to recommend professionals during onboarding, but it was optional. The theory was that every satisfied client could carry a good professional into another client's circle. A useful recommendation would generate a booking. A booking would create a working pair. That pair might invite more clients and professionals.
There was a catch hidden in the ordinary life of the customer. How often do you look for a new accountant, tutor or massage therapist? In his 2019 retrospective, Belsky estimated that the acute need might arise monthly at most, often quarterly. Prefer's delightful moment - finding the right person through a friend - happened too seldom. A consumer had little urgency to download the app, import contacts and recommend providers today for a problem they might have in three months. When the problem arrived, the group text was already waiting.
“You can work alone, but you are not alone.”Prefer's line for independent professionals, recalled by Scott Belsky
The message left the app
The other obstacle was more mundane and more revealing. Independent professionals and clients were used to coordinating by text. Prefer needed a place for messages, schedules, bookings and payments; many people saw no reason to move a relationship that already worked in SMS. Belsky later argued that a network can ask people to learn a new behavior only when the new way is dramatically better. Prefer's internal workflow had not earned that change. Missing pieces, including early calendar synchronization, made the move harder.
Once a client and professional began working together through Prefer, they often continued. That made the "working pair" a promising unit to measure, more useful than app downloads alone. But the company could not form enough pairs without labor-intensive help. Belsky concluded that productivity tools could support retention, yet were weak reasons for someone to join in the first place. It is a distinction many software businesses learn only after polishing the wrong screen.
Prefer also faced a structural choice. A typical marketplace starts narrow so buyers can find plenty of relevant sellers. Prefer had reason to start wide: a massage therapist might hesitate to bring clients into a marketplace packed with rival therapists, but could invite them into a network spanning tutors, chefs and accountants. The broad approach helped the referral logic and diluted density in each category. The company needed a network both large enough to be useful and intimate enough to feel safe.
What the pivots spent
Belsky described at least three major pivots before Prefer closed. He thought some changes came too quickly. The team learned that users wanted quite different things: some consumers wanted discovery, some professionals only referrals, others a better way to serve existing clients. A single onboarding path struggled to make sense to all of them. The "first mile" of the product, where a new user must understand the reward before doing work, had not improved fast enough for the range of people the network needed.
The cost was not published as a tidy dollar figure. The Series A amount was undisclosed, as were Prefer's revenue and valuation. The more legible cost was human: each pivot asked the same team to discard code, design and plans while carrying yesterday's doubts into tomorrow's test. Belsky said energy and patience were noticeably thinner after the third major reset. A pivot is a fresh strategy; it is rarely a fresh nervous system.
One part did live beyond the app. Prefer's member-led Soloist Collective connected independent professionals for peer support, shared practices, informal referrals and gatherings. Belsky wrote that it continued growing organically after the product had gone. The community suggests that Prefer had diagnosed a real social problem even as its consumer app struggled to become a routine solution.
The lesson worth stealing
A reader building a service marketplace can copy several experiments without copying Prefer's fate. First, identify the smallest real relationship that proves value - in this case, a client and professional who work together again - and track it before celebrating signups. Second, test how often the customer's problem occurs. If it appears once a quarter, the product needs a route back into daily life or a distribution channel that catches people at the moment of need. Third, let people use the communication habits they already have until a new workflow is clearly better. Finally, ask whether the community is the product, or whether the software is merely a meeting place for one.
Prefer's model might have worked differently in a category where customers repeatedly search for new providers, or where booking and administration were painful enough to pull existing pairs into one system. Belsky's retrospective does not offer a perfect counterfactual. It offers something more useful: the stubborn fact that liking a recommendation is not the same as opening an app. Trust can win a decision. Habit wins a market.